← Back to context

Comment by zuzululu

3 days ago

This moment is coming for software developers too

Yeah almost certainly, especially the ones who made a career out of "copypaste from StackOverflow", which is most engineers.

But even the good engineers should likely be a little worried.

  • why would it be different for other people if you already said senior level is not writing code but planning things out?

    is there something about planning that LLMs cannot do being your crux of the argument?

    what do you believe about your jobs or function that you think will be immune from AI replacing you?

    If anything it seems your role is not that dissimilar to those translating languages or business requirements.

    I am struggling to see what it is about this planning you do that cannot be done by AI because it seems to me thats not where the moat is rather I find the middle man jobs to be the most vulnerable to AI immediately much more than people writing code.

    Because at least someone is watching the outputs from AI and understands the code and can communicate it easily back to the stakeholders without the middle man gate keeping and applying their "taste".

    I have a feeling that anyone in your shoes is going to be working with code soon or they won't have much to offer anymore to the business. A stakeholder could easily replace the middle layer with AI and even as a business owner myself I do not see any need to add any more humans at the layer anymore unless they write code.

More specifically, it is coming for coders. If you make your living by banging out lines of code all day, then you may want to be looking at adjusting your career trajectory. But if that is your job, you are either very junior, or a bit foolish for getting into that situation.

  • so what is software developer doing if writing code is not part of their job

    I don't see how not writing code is being offered as a moat, it seems like that is just translating business/stakeholder requirements to architecture/biz processes which is exactly the type of low hanging fruit that AI will capture first

    or was it your point that the position sits closer to the stakeholders (relatively compared to those lifting) thus immune from replacement by AI

    or is your argument that your taste is exquisite that no AI will be able to match it like it already has with software so far and it will not improve beyond the current state

    • I kind of view it this way. Yes, non-technical people can prompt and write code. But technical people can certainly also do the job of product people. So then, who would you want to do the end-to-end? A SWE, or a product person? Probably a SWE.

      As software engineers, our job will expand horizontally. We will shift left, and right. But that’s really fine, because I’ve found SWE can be really good at that.

      They’re good at requirements engineering. They’re good at quality assurance. They’re good at technical support. So, why not pay a SWE to be that person?

      Or, at least, some SWEs are good at that. The ones that aren’t will struggle I think.

    • If you get to senior level then most of your job probably is not writing code, but planning things out. The code is largely an implementation detail.

      At least that's how it was for me, maybe other peoples' careers are different.

      7 replies →

    • Same thing architects do if drawing lines gets automated: architecture.

      Would you trust living in a high rise designed by AI?

      Designing a system that survives production is the job.

I think this collapses a global, complex heirarchy of software engineering workers into a single monolith and serves only to advertise for frontier LLM providers. the point where you no longer need engineers is not going to be reached by making LLMs better and better.

I think there is going to be a long time before all of the obscure knowledge of a decent software developer can be completely replaced by AI. Though the job is going to change beyond recognition. It already has in many ways.

But not before a huge crash in optimism about their capabilities. Specifically wrt accuracy, reliability, efficiency, and organization/architecture.